Sequential interleukin 3 and granulocyte-macrophage–colony stimulating factor therapy in patients with bone marrow failure with long-term follow-up of responses

Abstract

BACKGROUND

Interleukin-3 (IL-3) and granulocyte-macrophage–colony stimulating factor (GM-CSF) have synergistic, hematopoietic growth-promoting activity in preclinical studies. Because of the paucity of effective therapies for patients with chronic bone marrow failure states, the authors studied the biologic activity of sequential IL-3/GM-CSF in such patients.

METHODS

IL-3 was given subcutaneously for 5 days (at escalating doses of 0.15 μg/kg, 0.3 μg/kg, 0.6 μg/kg, 1.2 μg/kg, 2.5 μg/kg, 5.0 μg/kg, 10.0 μg/kg, or 15.0 μg/kg per day), and GM-CSF for was given subcutaneously for 9 days (at a dose of 5 μg/kg per day; Phase I 3 + 3 design) followed by 14 days of rest (total, 2 courses), then maintenance therapy.

RESULTS

The majority of 38 evaluable patients had aplastic anemia or myelodysplastic syndrome. Most patients (79%) had neutrophil responses. Ten patients (26%), all of whom were treated with IL-3 doses ≥ 1.2 μg/kg per day, had platelet responses, with a median increase of 132 × 10⁹/L (range, 41–180 × 10⁹/L) over baseline in responders. Six patients (16%) had trilineage recovery, which could be durable (the longest ongoing at 6.5 years after therapy completion). The most common toxicities were low-grade fever, headache, and fatigue. The maximum tolerated doses were IL-3 at 10 μg/kg per day and GM-CSF at 5 μg/kg per day.

CONCLUSIONS

Sequential IL-3/GM-CSF effectively raised blood counts in some patients with bone marrow failure at doses that were tolerated well. These results indicate that early-acting growth factors can induce durable, multilineage responses in a subset of individuals with bone marrow failure. Cancer 2003. © 2003 American Cancer Society.

Bone marrow failure includes aplastic anemia (AA), myelodysplastic syndrome (MDS), and prolonged cytopenia caused iatrogenically by chemotherapy, radiotherapy, or bone marrow transplantation (BMT). Allogeneic BMT from a human leukemic antigen (HLA)-matched sibling donor provides curative therapy for patients with AA but is limited by donor availability.1–6 Immunosuppression with antithymocyte globulin (ATG) and cyclosporine A is an effective alternative treatment.7, 8 Hematologic improvement generally occurs within 3 months, and up to 70% of patients respond.8 New therapeutic options are needed for patients who have failed immunosuppressive therapy and who do not have a sibling donor.

Supportive care is a common management strategy for patients with MDS who have early-stage disease.9, 10 Other therapies include allogeneic BMT, conventional and intensive chemotherapy, and growth factor treatment. Allogeneic BMT for MDS11–16 is limited by treatment-related morbidity/mortality, older age at diagnosis, and lack of sibling donors. The other therapies have not demonstrated an impact on survival.17–22

Diverse growth factors have been used in clinical trials for patients with bone marrow failure of various etiologies.23–26 Granulocyte-macrophage–colony stimulating factor (GM-CSF) and granulocyte-colony stimulating factor (G-CSF) produce neutrophil responses in the majority of patients. Recent studies suggest that low doses of interleukin 11 (IL-11) may induce platelet responses in 35–40% of patients.27 Similarly, preliminary trials suggest that stem cell factor (SCF) may induce platelet and multilineage responses in patients with AA.28, 29 IL-3 alone, however, has very limited activity.30, 31 Overall, enthusiasm for clinical development of growth factors in bone marrow failure states has been dampened by the paucity of reports of multilineage and/or durable responses.

Ultimately, the alleviation of bone marrow failure by growth factors most likely requires combinations tailored to the needs of the disease state and perhaps to the individuals themselves. IL-3 and GM-CSF are synergistic in vitro and in animal models, especially using sequential administration.32, 33 We performed a Phase I study of a fixed, standard dose of GM-CSF (5 μg/kg per day) given for 9 days preceded by 5 days of escalating doses of IL-3. Our results indicate that this combination can be administered safely and that durable platelet and multilineage responses are achievable in a subset of patients with bone marrow failure. The possibility of durable multilineage recovery with growth factors alone suggests that further trials of a broad spectrum of early-acting growth factors or their analogs in bone marrow failure merit exploration.

MATERIALS AND METHODS

Eligibility

Patients had bone marrow failure verified by bone marrow examination. Neutrophil counts were ≤ 1.5 × 10³/μL, and platelet counts were ≤ 50 × 10⁹/L. Karnofsky performance status (Eastern Cooperative Oncology Group; 0–2)34 and renal and hepatic function (creatinine ≤ 2 mg/dL; bilirubin ≤ 1.5 mg/dL) were adequate. Patients with AA were to be offered primary therapy with ATG, and at least 3 months must have elapsed since they had received such therapy without evidence of response or with loss of initial response. All patients were ineligible for matched-sibling BMT. Written informed consent was obtained from patients in accordance with M. D. Anderson Cancer Center Surveillance Committee guidelines.

Treatment Plan

Recombinant human IL-3 (Sandoz-Schering, Basel, Switzerland) at dose levels of 0.15 μg/kg, 0.3 μg/kg, 0.6 μg/kg, 1.2 μg/kg, 2.5 μg/kg, 5.0 μg/kg, 10.0 μg/kg, and 15 μg/kg per day subcutaneously was given for 5 days followed by 9 days of subcutaneous GM-CSF (Sandoz-Schering) 5 μg/kg per day. Then, there was a 2-week rest period followed by a second 14-day course of IL-3/GM-CSF at the identical dose and schedule. At least three patients were entered at each dose level. The maximum tolerated dose (MTD) was defined as the dose level below that at which one-third of patients had National Cancer Institute Common Toxicity Criteria (NCI CTC) Grade 3 or 4 toxicity. For patients with Grade 3 toxicity, the dose of either GM-CSF or IL-3 (depending on which drug was being given at the time) could be reduced by 50% (after withholding the drug until recovery from the toxicity). Patients with Grade 4 toxicity were removed from the study.

All patients who had any evidence of hematologic improvement after the induction phase (2 courses; 8 weeks) could be given maintenance therapy using a similar dose. The duration of the rest period between courses could be adjusted to keep blood counts in the physiologic range.

Response Criteria

All responses were documented on at least three consecutive measurements without transfusion. Baseline counts were the median of the 3 untransfused counts available within the 2 weeks before the initiation of therapy. Neutrophil response was defined as absolute neutrophil count (ANC) doubling and reaching a level ≥ 1 × 10³/μL. Platelet response was defined as doubling and reaching a level ≥ 50 × 10⁹/L (for patients with baseline platelet counts of 20–50 × 10⁹/L) or tripling and reaching a level ≥ 20 × 10⁹/L (for patients with baseline platelet levels < 20 × 10⁹/L). Hemoglobin response was defined as an increase ≥ 2 g/dL or transfusion independence in transfusion dependent patients.

RESULTS

Patient Characteristics

Thirty-nine patients were registered on the trial, all of whom were evaluable for toxicity. Thirty-eight patients were evaluable for response (Table 1). Most patients had AA (n = 20 patients) or MDS (n = 12 patients). The 5 patients with iatrogenic bone marrow failure were 2 months, 4 months, 15 months, and 18 months after successful chemotherapy for testicular carcinoma, lymphoma (n = 2 patients), and B-cell acute lymphoblastic leukemia (ALL), respectively, and 5 months after the second autologous BMT for lymphoma. There were a wide variety of prior therapies (Table 2). Thirty-three patients (85%) were dependent on red blood cell transfusions, and 24 patients (62%) were dependent on platelet transfusions.

Table 1. Patient Characteristics

Characteristic	No. of patients
AA: aplastic anemia; RA: refractory anemia; RAEB: RA with excess blasts; RAEBT: RAEB in transformation. a One patient with hairy cell leukemia was registered in error and not evaluable for response. b Bone marrow failure was due to chemotherapy and/or bone marrow transplantation for testicular carcinoma (n = 1 patient), lymphoma (n = 3 patients), and B-cell acute lymphoblastic leukemia (n = 1 patients). Patients were from 2 months to 18 months after therapy without bone marrow recovery.
Patients registered	39
Patients assessable for toxicity	39
Patients assessable for response	38
Reasons patient not evaluable
Registered in errora	1
Age (yrs)
Median	56
Range	20–78
Gender
Male	23
Female	16
Diagnosis
AA	19
RA	5
RAEB	6
RAEBT	1
Bone marrow failureb	5
Myelofibrosis	2
Hairy cell leukemiaa	1
Karyotype
Diploid	24
Insufficient metaphases	3
5q−	1
7q−, 22q+	1
Monosomy 9	1
−8	1
45x, −y	1
del 11q14	1
Others	5

Table 2. Prior Therapy for Bone Marrow Failure

Prior therapy	No. of patients (%)
ATG: antithymocyte globulin; GM-CSF: granulocyte-macrophage–colony stimulating factor; IVIG: intravenous immunoglobulin; M-CSF: macrophage-colony stimulating factor; Ara-C: cytosine arabinoside; IFN: interferon; IL-3: interleukin 3; BMT: bone marrow transplantation; SCF: stem cell factor; VP-16: etoposide; PIXY: genetically engineered interleukin 3/granulocyte-macrophage–colony-stimulating factor hybrid molecule.
None	12 (30.8)
Steroid	12 (30.8)
Cyclosporine	8 (20.5)
ATG	8 (20.5)
Erythropoietin	7 (17.9)
Danazol	4 (10.3)
GM-CSF	4 (10.3)
Halotestin	3 (7.7)
Neupogen	3 (7.7)
IVIG	2 (5.1)
M-CSF	2 (5.1)
Ara-C	2 (5.1)
IFN	2 (5.1)
IL-3	2 (5.1)
BMT	1 (2.6)
IL-4	1 (2.6)
SCF	1 (2.6)
Androgen	1 (2.6)
VP-16	1 (2.6)
PIXY	1 (2.6)

Responses

Ten patients (26%) had a platelet response, 6 of whom (16% of total) showed trilineage responses (Tables 3, 4). It is noteworthy that all patients who had a hemoglobin response eventually attained a trilineage response.

Table 3. Responses by Dose Level

IL-3 dose (μg/kg/day)	0.15	0.3	0.6	1.2	2.5	5.0	10.0	15.0	Total (%)
IL-3: interleukin 3; GM-CSF: granulocyte–macrophage-colony stimulating factor. a One patient was unevaluable for response due to an incorrect diagnosis (hairy cell leukemia). b Platelet and neutrophil responses without hemoglobin increase.
GM-CSF dose (μg/kg/day)	5.0	5.0	5.0	5.0	5.0	5.0	5.0	5.0	—
No. of patients treated	3	3	3	6	3	7	8	6	39
No. of patients evaluable for responsea	3	3	3	6	2*	7	8	6	38
Platelet response	0	0	0	2	1	3	2	2	10 (26.3)
Hemoglobin response	0	0	0	1	0	2	2	1	6 (15.8)
Neutrophil response	2	1	3	3	2	6	7	6	30 (78.9)
Bilineage response onlyb	0	0	0	1	1	1	0	1	4 (10.5)
Trilineage responses	0	0	0	1	0	2	2	1	6 (15.8)

Table 4. Responses by Diagnosis

Diagnosis	Total no. of patients	No. of responses (%)
		Platelet	Bilineage onlya	Trilineage
AA: aplastic anemia; MDS: myleodysplastic syndrome; BMF: bone marrow failure; BMT: bone marrow transplantation. a Of 10 patients who had bilineage responses, all had platelet and neutrophil responses. The percentage response was calculated based on the total number of patients with that diagnosis. b Platelet responders were 4 months, 5 months, and 18 months postchemotherapy or post-BMT at the time of study entry.
AA	19	5 (26)	1 (5)	4 (21)
MDS	12	2 (17)	2 (17)	0 (0)
BMF postchemotherapy or BMTb	5	3 (60)	1 (20)	2 (40)
Myelofibrosis	2	0 (0)	0 (0)	0 (0)
Total	38	10 (26)	4 (11)	6 (16)

Responses were dose-related: No platelet responses were seen at IL-3 doses < 1.2 μg/kg per day. At IL-3 doses ≥ 1.2 μg/kg per day, 10 of 29 patients (34%) had platelet responses, and 6 of 29 patients (21%) had trilineage responses (AA; n = 4 patients) and bone marrow failure 5 months after autologous transplantation (n = 1 patient) or 18 months after chemotherapy (n = 1 patient). All patients with AA who responded had not received prior therapy with ATG within the 6 months before entering the study.

Three of the trilineage responders have ongoing normal or near-normal counts 4.0–6.5 years after the completion of therapy. One trilineage responder lost his response during the 10th month of treatment, when his ALL recurred. The other 2 patients lost their responses soon after they were taken off therapy at 3 months (due to urticaria) and at 2.5 years (due to recurrent malignancy).

Four patients demonstrated responses in absolute neutrophils and platelets without a response in hemoglobin (bilineage response) (Tables 3, 4). A total of 30 patients (79%) had a neutrophil response (Table 3).

Responses in neutrophils generally were observed within 1–3 days after the initiation of therapy. Initial platelet or hemoglobin increases, however, often were evident only after about 1–3 months of treatment. However, 2 patients showed a rapid rise in platelet counts within 1–2 weeks after the start of treatment. In many patients, responses in these lineages were gradual, and peak responses were attained ≥ 1 year after therapy initiation. Two patients, both of whom had myelofibrosis, demonstrated a rapid increase in spleen size suggestive of progressive disease.

Platelet Responders

The courses of six patients with platelet responses, five of whom had trilineage recoveries, are reported in depth. These cases illustrate the variability in time to response, the durability of response, as well as the temporal connection between courses of therapy and changes in blood counts. They also demonstrate that the increases in blood counts observed considerably exceeded the minimal criteria for response, with some patients achieving normal or near-normal blood counts.

Case 1.

Patient 28 was a woman age 21 years who had a diagnosis of AA (8 months' duration). Her karyotype was diploid. Prior therapy with ATG caused severe toxicity without response. Cyclosporine was discontinued because of significant hepatic toxicities. The patient had no HLA-matched sibling for BMT. IL-3/GM-CSF was initiated 7 months after ATG and 6 months after cyclosporine treatments were withdrawn. Her baseline platelet count was 31 × 10⁹/L, hemoglobin was 10 g/dL, and ANC was 1.0 × 10³/μL. She was treated with IL-3 at a dose of 10 μg/kg per day and GM-CSF for a total of 27 courses without significant toxicity. A definitive rise in platelet count was not noted until after the sixth course, at which time the rest period between courses had been decreased from 2 weeks to 1 week (Fig. 1). The peak platelet count was 174 × 10⁹/L, the peak ANC was 15.4 × 10³/μL, and the peak hemoglobin was 14.9 g/dL (Fig. 1). Bone marrow biopsies showed increased cellularity from 20% to 85% during treatment. There was no change in the percentage of blasts or the degree of dysplasia, and cytogenetics remained diploid. At the last follow-up, the patient had been off therapy for > 5 years and remained in remission with normal blood counts in all 3 lineages. She had delivered a healthy child.

Figure 1. Platelet (Plt) and hemoglobin (Hb) response of Patient 28. Courses of interleukin-3 (IL-3)/granulocyte-macrophage–colony stimulating factor (GM-CSF) (n = 27 courses) are depicted in the bars below the graph (IL-3 dose, 10 μg/kg per day; GM-CSF dose, 5 μg/kg per day). Normalization of neutrophils was observed (data not shown).

Case 2.

Patient 30 was a man age 54 years who had been diagnosed with B-cell ALL 3 years earlier. He had diploid cytogenetics. He was treated with multiagent chemotherapy (seven courses) with a complete response. Eighteen months after chemotherapy, he remained severely cytopenic and suffered from gastrointestinal bleeding and sepsis, requiring intensive care unit management. Bone marrow aspiration and biopsies performed on three occasions showed no evidence of ALL. He was treated with intravenous immunoglobulin, danazol, and steroids without response. The patient received an IL-3 dose of 10 μg/kg per day followed by GM-CSF at 5 μg/kg per day. His baseline platelet counts were 1–2 × 10⁹/L, and he was dependent on red blood cell transfusions. His ANC was 0.2 × 10³/μL. The platelets started to rise on Day 8 of the first course, fluctuated during early treatment, and reached a peak of 141 × 10⁹/L. Hemoglobin and neutrophil counts also rose (Fig. 2). He received a total of 10 courses (10 months of therapy) without significant side effects. Unfortunately, ALL recurrence was then noted, and chemotherapy for ALL was restarted.

Figure 2. Platelet (Plt) and hemoglobin (Hb) response of Patient 30. Courses of interleukin-3 (IL-3)/granulocyte-macrophage–colony stimulating factor (GM-CSF) (n = 10 courses) are depicted in the bars below the graph (IL-3 dose, 10 μg/kg per day; GM-CSF dose, 5 μg/kg per day). Normalization of neutrophils was observed (data not shown). RBC: red blood cells.

Case 3.

Patient 15 was a woman age 20 years with a diagnosis of AA of 10 months' duration. She had diploid cytogenetics, and her baseline platelet count was 20 × 10⁹/L, hemoglobin was 10.1 g/dL, and ANC was 1.5 × 10³/μL. The patient had no matched-sibling donor for BMT, and she refused immunosuppressive therapies. She was treated with IL-3 at a dose of 1.2 μg/kg per day and GM-CSF at 5μg/kg per day for a total 22 courses without significant toxicity. The platelet count peaked at 166 × 10⁹/L, the hemoglobin ranged from 11.0 g/dL to 12.6 g/dL, and the ANC was ≈ 5 × 10³/μL (Fig. 3). The platelet and neutrophil response is ongoing at 4 years after the last course. Her anemia worsened after two successful pregnancies, during and after which she refused iron supplementation. The bone marrow cellularity increased from 10% (baseline) to 60% during treatment, and there was no significant change in blast counts or in karyotype.

Figure 3. Platelet (Plt) and hemoglobin (Hb) response of Patient 15. Courses of interleukin-3 (IL-3)/granulocyte-macrophage–colony stimulating factor (GM-CSF) (n = 22 courses) are depicted in the bars below the graph (IL-3 dose, 1.2 μg/kg per day; GM-CSF dose, 5 μg/kg per day). Thin arrows represent platelet transfusions. Normalization of neutrophils was observed (data not shown).

Case 4.

Patient 23 was a man age 49 years who had been diagnosed with large cell lymphoma 2.5 years earlier. He was treated with multiple forms of chemotherapy and underwent two autologous transplantations because of repeated recurrences. The patient achieved complete disease remission but suffered prolonged pancytopenia 5 months after his second transplantation. At that time, he was enrolled into the protocol. His baseline platelet count was 5 × 10⁹/L, he was red blood cell transfusion dependent, and his ANC was 0.8 × 10³/μL. Baseline bone marrow blasts were 0.0–1.5%, cytogenetics were diploid and bone marrow cellularity was 5% with megakaryocytic hypoplasia. He was treated with IL-3 at a dose of 5 μg/kg per day and GM-CSF at 5 μg/kg per day for a total of 16 courses with a multilineage response. No significant side effects were noted. After treatment, the blood counts normalized (Fig. 4). At the time of last follow-up, the patient has been followed for 6.5 years off therapy. Lymphoma remained in complete remission, and his blood counts were normal. The bone marrow cellularity increased from 5% (baseline) to 50% during treatment, and there was no significant change in blast counts or evidence of clonal evolution.

Figure 4. Platelet (Plt) and hemoglobin (Hb) response of Patient 23. Courses of interleukin-3 (IL-3)/granulocyte-macrophage–colony stimulating factor (GM-CSF) (n = 16 courses) are depicted in the bars below the graph (IL-3 dose, 5 μg/kg per day; GM-CSF dose, 5 μg/kg per day). Arrows represent red blood cell (RBC) and platelet transfusions, respectively. Normalization of neutrophils was observed (data not shown).

Case 5.

Patient 20 was a man age 69 years who had been diagnosed with large cell lymphoma 2 years earlier. He was treated with alternating triple chemotherapy with complete remission. The patient developed a pelvic lymphoma recurrence 16 months later and was treated with radiation therapy to the pelvis and then received mantle field radiation. One month after the completion of radiation therapy, the patient developed severe thrombocytopenia. A workup did not reveal evidence of recurrent lymphoma. Bone marrow biopsy showed 20% cellularity, 2% blasts, megakaryocytic hypoplasia, and normal diploid cytogenetics. His baseline platelet count was 4 × 10⁹/L, baseline hemoglobin was 9.2 g/dL, and ANC was 4.8 × 10³/μL. He was enrolled in the study 4 months after radiation was completed and was treated with IL-3 at 5 μg/kg per day and GM-CSF at 5μg/kg per day. His platelet count began to increase, stabilized above 20 × 10⁹/L during the second course, and rose to 45 × 10⁹/L during the third course. The patient had a history of atrial fibrillation 5 years earlier and developed a recurrent episode during the first course of treatment. The atrial fibrillation resolved with digoxin despite continuation of treatment. For personal reasons, the patient declined treatment after the third course. His platelet count subsequently decreased to baseline over a 2-month period (Fig. 5).

Figure 5. Platelet (Plt) response of Patient 20. Courses of interleukin-3 (IL-3)/granulocyte-macrophage–colony stimulating factor (GM-CSF) (n = 3 courses) are depicted in the bars below the graph (IL-3 dose, 5 μg/kg per day; GM-CSF dose, 5 μg/kg per day). Arrows represent platelet transfusions.

Case 6.

Patient 29 was a man age 78 years man with transitional cell carcinoma of the bladder diagnosed initially 10 years earlier and treated with surgery and thiotepa. Seven months after his last dose of thiotepa, the patient developed pancytopenia. Bone marrow aspiration and biopsy performed at time of referral showed 0.5% blasts, 15% cellularity, and no significant dysplasia. He had diploid cytogenetics. A prior course of ATG 8 months earlier did not result in a response. His diagnosis was consistent with AA. He was enrolled in the protocol with an IL-3 dose of 5 μg/kg per day and a GM-CSF dose of 5 μg/kg per day. He had a rapid rise in platelet counts after the first course, with a platelet peak of 220 × 10⁹/L (baseline, 40 × 109/L), and his hemoglobin increased from a baseline level of 7.8 g/dL to 12.0 g/dL (Fig. 6). In addition, his neutrophil counts increased from 0.4 × 10³/μL to 2.3 × 10³/μL. The response lasted for about 2 months after the treatment, then the counts gradually decreased. He received a total of 11 courses. The rest period between courses was prolonged to approximately 8–10 weeks because of the durable elevation of blood counts after each course. Because of headache and chest pain during the 4th course, the dose of IL-3 was decreased to 2.5 μg/kg per day. Tolerance was excellent, and the IL-3 dose was then increased to 3.75 μg/kg per day. After the 11th course, he declined further therapy because of intercurrent medical problems (recurrent bladder carcinoma). The bone marrow cellularity increased from 15% to 40% without significant changes in blast counts.

Figure 6. Platelet (Plt) and hemoglobin (Hb) responses of Patient 29. Courses of interleukin-3 (IL-3)/granulocyte-macrophage–colony stimulating factor (GM-CSF) (n = 11 courses) are depicted in the bars below the graph (IL-3 dose, 5 μg/kg per day; GM-CSF dose, 5 μg/kg per day). Normalization of neutrophils was observed (data not shown).

Side Effects

Overall, sequential IL-3/GM-CSF combination treatment was tolerated well (Table 5). The most common side effects were Grade 1 and 2 fever, headache, fatigue, myalgia, nausea, emesis, diarrhea, dizziness, and local reaction. Grade 3 or 4 toxicities included pulmonary symptoms (edema, wheezing, and/or pleuritic pain; n = 3 patients), headache (n = 2), cardiomegaly and congestive heart failure (n = 1 patient), nausea and emesis (n = 1 patient), subconjunctival hemorrhage (n = 1 patient), and subarachnoid hemorrhage (n = 1 patient). The latter patient had severe AA (baseline platelet count, 6 × 10⁹/L) and a history of subarachnoid hemorrhage twice before enrollment and died on Day 14 of the first course. Seven patients (18%) had no side effects.

Table 5. Toxicity of Interleukin 3/Granulocyte-Macrophage–Colony Stimulating Factor

Toxicity	No. of patients (%)	No. with Grade 3/4 tumors (%)
a Pulmonary symptoms include edema, wheezing, and/or pleuritic chest pain. b Cardiac toxicity consisted of cardiomegaly with symptoms of congestive heart failure and/or arrhythmia.
Fever	11 (28.2)	—
Headache	8 (20.5)	2 (5.1)
Fatigue	7 (17.9)	—
Pulmonarya	6 (15.4)	3 (7.7)
Myalgia	5 (12.8)	—
Nausea/emesis	3 (7.7)	1 (2.5)
Subconjunctival bleed	2 (5.1)	1 (2.5)
Diarrhea	2 (5.1)	—
Dizziness	2 (5.1)	—
Local reaction	2 (5.1)	—
Cardiacb	1 (2.5)	1 (2.5)
Intracranial hemorrhage	1 (2.5)	1 (2.5)
Neutropenic infection	1 (2.5)	—
Rash	1 (2.5)	—
Urticaria	1 (2.5)	—

Maximum Tolerated Dose

At a dose of 15 μg/kg per day of IL-3 followed by GM-CSF (5 μg/kg per day), 4 of 6 patients required dose reductions during the first 2 courses because of side effects: headache (n = 2 patients), nausea/emesis (n = 1 patient), and subconjunctival bleeding (n = 1 patient). In addition, one patient developed cardiomegaly necessitating dose reduction during the third course. At a dose of 10 μg/kg per day of IL-3, only 2 of 8 patients had Grade 3–4 toxicity (pulmonary edema; n = 2 patients). Therefore, the MTD was 10 μg/kg per day for IL-3 when followed by GM-CSF doses of 5 μg/kg per day.

DISCUSSION

Myeloid growth factors such as G-CSF have demonstrated remarkable salutary effects in a variety of clinical settings. However, a similarly efficacious platelet or multilineage growth factor has not been identified. Even so, several cytokines are able to increase platelet counts to various extents. Indeed, IL-11 has been approved in the U.S. for the attenuation of postchemotherapy thrombocytopenia. Thrombopoietin also may alleviate thrombocytopenia after chemotherapy.35, 36 In the bone marrow failure setting, clinical effects have been both less studied and less pronounced. For instance, a beneficial effect of thrombopoietin in graft failure could not be demonstrated, although this may have been the result of the intermittent dosing schedule used.37 IL-3 administered as a single agent also had little impact on the bone marrow failure state.31 Combinations of erythropoietin together with myeloid growth factors usually have demonstrated only bilineage responses, although occasional platelet recovery is noted.38 Still, IL-11 may increase platelet counts in a subset of patients with MDS or AA,27 and durable multilineage recovery has been observed in some patients with AA who were treated with SCF.29 Taken together, these results suggest that both platelet and multilineage responses are achievable in bone marrow failure but that additional molecules warrant investigation.

IL-3 stimulates multilineage hematopoietic progenitors in both in vitro and in vivo studies,33, 39–41 and endogenous IL-3 levels may be reduced in patients with AA.42 In clinical trials, IL-3 has induced sustained remission in patients with Diamond–Blackfan anemia,43, 44 albeit rarely. This molecule induces only modest hematopoietic effects in a minority of patients with bone marrow failure.31 IL-3 is tolerated well at doses up to 25 μ g/kg per day in this setting.31 Tolerance to an identical IL-3 compound produced by the same source, however, was much less after BMT with an MTD of 2 μg/kg per day.45 The reason for these differences is unclear but may be due to synergy with endogenous cytokine release after transplantation. Similar to the results with IL-3 alone, our current study demonstrates that, in patients with bone marrow failure, sequential IL-3/GM-CSF is tolerated well with an MTD for IL-3 of 10 μg/kg per day.

GM-CSF is known best for its ability to increase neutrophil counts, although it also can stimulate the growth of multilineage progenitors.28, 46 This molecule has been used as a single agent in patients with bone marrow failure; and increases in granulocytes, monocytes, and eosinophils have been demonstrated.23, 47 Both platelet increases and platelet decreases occur, albeit infrequently.48

IL-3 and GM-CSF have overlapping but distinct activities.46 In addition, in vitro data demonstrate that a combination of IL-3 and GM-CSF is more potent than either used alone.32, 49, 50 Sequential administration of IL-3 followed by GM-CSF resulted in a shortened duration of neutropenia and attenuation of thrombocytopenia and anemia in patients who were treated with high-dose chemotherapy for lymphoma or breast carcinoma with or without autologous BMT.51–54 However, PIXY-321 (an IL-3/GM-CSF fusion protein) had little impact on platelet transfusion requirements after transplantation.55 Sequential IL-3/GM-CSF (IL-3 doses ≤ 1 μg/kg per day) also has been tested in a small group of patients with MDS (n = 9 patients).56 A clinically relevant response was seen in only one patient with thrombocytopenia. The infrequency of response at this dose level is consistent with our data. We did not observe platelet responses in any of the 9 patients treated at doses of IL-3 < 1.2 μg/kg per day; whereas, at doses ≥ 1.2 μg/kg per day of IL-3, 34% of patients demonstrated platelet increases. In addition, a subset of our patients (n = 6 patients; 16% of the total group and 21% of patients treated with IL-3 doses ≥ 1.2 μg/kg per day) achieved responses in all 3 lineages. In 3 of the trilineage responders, responses are ongoing 4.0–6.5 years after the treatment was discontinued. A see-saw pattern of platelet increases (corresponding with the on treatment and off treatment periods) was seen often, with platelet counts rising during the first week of the rest period between courses and then dropping during the second week but with both phases (peaks and nadirs) showing a slow but consistent increase with time (Figs. 1, 3, 6). Of special interest is the slow onset of platelet responses in many patients (1 week to 3 months). Furthermore, peak responses in some individuals occurred at > 1 year after the initiation of treatment. A similar delay in platelet recovery has been seen in our studies with SCF in patients with AA29 and in patients with bone marrow failure treated with low-dose IL-11.27 These results suggest that the temporal response paradigm of G-CSF and GM-CSF, in which administration of growth factors is followed by a neutrophil response within 24 hours, may not be applicable when other, earlier acting growth factors are used. Alternatively, the fact that delayed platelet responses have been observed with several distinct growth factor regimens may indicate that the physiology and, hence, the time course of platelet recovery after administration of thrombopoietic agents is fundamentally different from that of neutrophils. These differences are underscored further by the fact that, in patients with bone marrow failure, when GM-CSF or G-CSF is withdrawn (even after prolonged treatment), neutrophils fall to baseline levels within days. In contrast, durable platelet responses and even trilineage responses have been observed after SCF29 and in the current study. Blood counts in patients (e.g., Case 5) who have received only a few courses of IL-3/GM-CSF may take ≥ 2 months to return to baseline (Fig. 5). Some responders treated for > 1–2 years appear to have long-term durable remissions (Figs. 1, 3, 4). These results suggest that the use of thrombopoietic or multilineage growth factors can be effective but that long-term administration may be necessary for optimal benefit.

In the current study, we demonstrated that, at doses that are well tolerated, the sequential administration of IL-3 followed by GM-CSF can raise platelet counts in patients with bone marrow failure. Furthermore, multilineage responses occur and may be durable. Because therapeutic options for these patients are limited, the results of the current study suggest that further exploration of early-acting and/or combinations of growth factors is warranted in the setting of bone marrow failure.